Logistics, market sizeand giantplantsin the early 20th century: a global view.

by Leslie Hannah

Professor of Economics, University of Tokyo;

Directeur d’Etudes Associé, Ecole des Hautes Etudes en Sciences Sociales, Paris.

Department of Economics

University of Tokyo,

7-3-1 Hongo,

Bunkyo-ku,

Tokyo 113-0033,

Japan.

620, Maison Suger

16-18 Rue Suger,

75006 Paris,

France.

This paper has been prepared to introduce a Round Table Discussion at the European Science Foundation’s “EUROCORES”Inventing Europe conference on 7-10 June 2007 in Rotterdam; for further details see the conference website at

Logistics, Market Size and Giant Plants in the early 20th Century: a Global View.

ABSTRACT

Around 1900, thebusinesses of developed Europe – transporting freight by a more advantageous mix of ships,trains and horses– encounteredlogistic barriers totrade lower than the tyranny of distance imposed on the sparsely populatedUnited States. Highly urbanized, economically integrated and compact northwest Europe was a market space larger than, and - factoring in other determinants besides its (low) tariffs - not less open to inter-country trade than the contemporary American market was to interstate trade. By the early twentieth century, theFirst European Integration enabled mines and factories– in small, as well as large, countries –to match the size of United States plants, where factor endowments, consumer demand or scale economies required that.

“Statistics of quantity, defective as they may be, are apt to form a surer basis of conclusions than statistics of value….It means little or nothing to say that the expense per train mile is a dollar in one country and eighty cents in another.”

Hadley, “Comparative Statistics of Railroad Service,” p. 36.

“We found there, as every attentive and expert traveller will find everywhere in the civilized world, some things better and some things less good than with us.”

1906 German Official Report on US Visit (Hoff and Schwabach, North

American Railroads, p. 412)

I am grateful to Lou Cain for critical comments on an earlier draft.

The United States, by the middle of the twentieth century,had achieved an, historically unprecedented, economy-wide, productivity advantage, with areal GDP larger than (and a real GDP per capita more than double)Western Europe’s.[1] The USA’s aggregatereal GDP was thennearlythree times the USSR’s, more than four times the UK’s and more than five times Germany’s. At that time the USA’s giantindustrial corporations were even more dominant: they outnumbered those of Europe by two or three to one.[2]Some investigators of the sources of American industry’s productivity leadthus naturallylinked it tothe unparalleled opportunity to achieve scale economies offered by the United States’ exceptionally large domestic market.[3] Subsequent historians have followed this lead. Alfred Chandler, for example, identified the smaller British market as one of the reasons why he believedits firms early in the century did not invest in large plants, distribution systems and managerial hierarchies on a sufficient scale, seeing the creation of a continent-wide market by the railroads as a major driver ofwhat he thought was the exceptional development of giant American corporations.[4] “New economic geographers” also emphasize that trade costs, scale economies, imperfect competition and knowledge spillovers interact to give large countries a disproportionate share of world industry.[5]

The interactions among market size, firm size, and productivity are complex and change over time. The United Statesovertook the real GDP of the largest European nation in the 1870s, butits manufacturing productivitymayhave forged aheadof Europe’sbefore it attained that scale advantage.[6] Moreover,market size is not simply determined by national boundaries: for many factoriesit will be smaller(a city or a region). It is affectednot only by tariff levels (of course, zero within the USA, but also quite low in Europe around 1900),but also, critically,by transport costs and factors such as linguistic or monetary homogeneity or urbanization. Britain and Germany(together, before World War One, having the same real GDP as the USA) – alongwith France, Switzerland, Belgium and the Netherlands- formed the largest compact urban market in the world and the massive trade flowsdriven from north-west Europe still accounted for most global manufacturing trade. This factor has been neglected by postwar historians, impressed byEurope’s1950-1975 catch-up, that was driven partly by reversing the destruction of people and propertyand tariff escalations that hadblighted its economic performance during itsvile military conflicts,cold wars,partitions, nostrifications and dictatorships of 1914-1945. This article argues that the leading European producers typically had access to at least as wide a market as American firms earlier in the twentieth century andthat they faced levels of cross-border market integrationnot dissimilar to today’s.There was simply no market scale reason why Europeanplantsaround 1900 could not be as large asAmerican ones; and, as it turns out,they were as large.

SHIPS, TRAINS AND HORSES

The choicebetween the two main logisticoptions of the nineteenth century depended partly on geography: there were no ships in Santa Fe and no trains to Hawaii. Yet, rail and water were substitutes on many long-haul routes, in global historical fact, as well as in Fogel’scounter-fact.[7] As today, in tonne-kilometre terms, water transportdominated long-haul, thoughby the early twentieth century the balance varied, with rail being exceptionally important for two large countries: the USA and Russia.[8] The data in Table 1 relate to transport use, not production: the latter would increase water’s share for the UK and Germany (large net exporters of shipping services) and lower it for Russia and the USA (the latter’s share of the world’s seagoing fleet had shrunk from 20% to 3% over a half-century, its imports and exports shifting overwhelmingly to foreign-flagged carriers; Russian overseas freight wasalso largely outsourced).

The reliability of the table falls off to the right: rail (and some domestic water) freight is reasonably hard data, the rest being crudely estimated froma coefficient relating the capacities of arriving and departing cargo-ships, voyage lengths and freight

Table 1. Freight Market Shares, ca. 1906.

Country Rail Water Transport Total Share

and ( Domestic and Inland. Cabotage. International. Freight of

Date International.) Market Rail

(units: billion tonne-km)

USA 1906 320 69 60 264 713 45%

“Europe”* 183 52 69 1532 1835 10%

of which:

Russia 1913 76.4 28.9 15.4 74.0 194.7 39%

Germany 1905 51.8 14.8 1.5 181.6 249.7 21%

UK 1910 22.1 1.1 33.5 585.7 642.4 3%

France 1906 18.2 5.1 1.9 247.5 272.7 7%

Belgium 1912 6.4 1.6 na 84.8 92.8 7%

Italy 1906 5.2 na 15.1 245.8 266.1 2%

Sweden 1913 3.2 na 1.2 112.3 116.7 3%

Japan 1908 3.0 na 25.0 244.9 270.2 1%

* “Europe” is thesum of the seven countries shown.

Sources: Cols. 1-3: Barger, Transportation Industries, pp. 184, 254-6; Bureau, Transportation, p.33 (USA, with Barger’s 1889 water traffic distance coefficients, derived from interwar data, applied to the 1906 regional tonnages); Scherer, USSR, p. 231 (Russia, post-1945 USSR borders, which approximate to those of the 1913 Empire excluding Finland, with 76% of Russian-ownedship tonne-km assumed to be in cabotage, following Williams, Freight Transportation, p. 12); Hoffmann, Wachstum, pp. 406-18, updated by Kunz, “Performance,” p.51 (Germany); Armstrong, “Role,” p176 (UK); Toutain, “Transports,” pp. 81, 158, 197 (France); Laffut, “Belgium,” pp. 217-218 (Belgium); Mitchell, International Historical Statistics, pp. 688-9 (Italy, with ratio of Italian cabotage derived from Schram Railways, p. 151 estimate for 1880/1; and Sweden, with cabotage from Krantz, “Competition,” p. 30); Minami, Railroads, p. 194 (Japan, with cabotage estimated by the author from indications in Ericson, Sound, pp.39-40, 397-8). Col. 4: Hoffmann’s estimates for the relationship between registered ship capacities, loads and voyage lengths are taken as the base to derive a coefficient, which is then applied to the national port data on the capacity of steamers entered and cleared with freight for the relevant year given in Anon, Statistical Abstract 1901/12, pp. 36-57 and Woytinsky, Welt, 5, p.77. International rail freight is apportioned in national statistics according to the distance traveled within each country (and cannot be disaggregated from domestic traffic; though see n. 12 below); for international sea freight I have apportioned 50% of voyage distances to destination and departure countries.

carried.[9] The precise quantities cannot be relied upon, though the orders of magnitude are broadly plausible. Islands or peninsulas - Britain, Italy, Japan –naturally used ships

prolifically, as did a littorally-settled continent like Australia (not shown).[10] Yet America’s use of rail was massively higher than even the continental expanse of Germany; only Russia approached America’s rail-dependence. The USA and Europe overall made roughly similar use of inland waterways (the Great Lakes and the Mississippi were more than a match for the Rhine and the Volga) and cabotage (sea routes between domestic ports). What gave Europe its massive lead in water transport was international shipping. In the case of intercontinental voyagesEurope was - by construction - an equal partner in US-Europe trade, but its global engagementwith Asia, Africa, Australasia and South America exceeded the USA’s.[11] Despitemany common land bordersbetween European countries, most cross-border tradewent by sea rather than rail.[12] Much of Europe’ssea-freightwas intra-European, and in that sense, treating Europe as one economic area,functionally equivalent to US cabotage.[13] TheWeltbűrger runninga Hamburg shipping enterprise– though not averse tonationalistically-motivated supports that came from Imperial Potsdam–knew his profits fundamentally came from serving customers in St Petersburg, Copenhagen, Cherbourg, London and elsewhere. In like cosmopolitan spirit, the English mariners’term “home trade” referred to voyages to and among nearby German, Dutch, Belgian and French ports between Hamburg and Brest. A large sample of crew agreements for 1863-1900 indicates that 43% of British steamer voyages were then within European waters, while a 1911 survey showed 21% of British shipping capacity operating within Europe.[14]By Hoffmann’s calculation, 43% of German-flagged capacityin 1913 was plying to European destinations, though he reckoned these exclusively European voyagesprobably accounted for only 11% of sea-freighttonne-km(oceangoing ships were larger, faster and spent most time at sea, while downtime in port, loading and unloading cargo, was the lot ofshort-haul captains).[15] Yet voyageshe designates as intercontinental also served intra-European trade: fast liners plying from Hamburg to New Yorkcalled at Cherbourg or Plymouth to pick up and set down small, high-value cargoes, as well as passengers;steamers from northern Europe that were Yokohama-bounddocked at Mediterranean ports. The slightly more numerousforeign ships entering and leaving German ports (mostEuropean-flagged) were also likely mainly engaged in intra-European trade, but are excluded fromHoffmann’scalculations.[16]Europe’s “domestic” sea freightalso had a higher value-to-volume ratio than the oceanic trades, perhaps reflecting the high share of manufactures and fuel carried.[17]

Ships had lower infrastructure costs (the sea was free) and also had the advantage of greater fuel-efficiency and lower terminal costs (cranes were ubiquitous in the main ports, and tramps carried their own lifting equipment for small ports, but manhandling – still widespread at all transport nodes -remainedcommon in rail trans-shipment).[18]Ships offered tonne-km freight rates only a half or a quarter of those of land transport, even, sometimes, as low as the one-seventh achieved by the modern development of supertankers and container shipping.[19] Yet trains could still compete if speed counted, though running at higher speeds was expensive in both modes.Britishfreight trainsaveraged around 32 kmph and Americanones around 17kmph, though they could, if required, go faster (one source mentions 77kph); at sea, trampsand tankers managed15-17 kmph andsome coastal liners 25 kmph, but 44kmph - the top speed of ships -was attained only bya few transatlantic liners.[20] Rail wasalsoused if there was no access by water, orit offered more direct routing. Some routes - Galveston to Key West, Duluth to Cleveland, Stockholm to St Petersburg, Newcastle to Hamburg, Trieste to Brindisi, Antwerp to Bilbao and Barcelona to Genoa – were more direct (andoften faster) by boat.Even circuitous routes, likeOdessa to London (2,308 km as the crow flies, but 6,326 km by steamer via the Bosporus, Mediterranean and Atlantic), were cheaper than the rail equivalent, especially for low-value traffic with low inventory costs.

Marine geography, and a semi-depleted continent’s global quest for raw materials that America found abundantly in its virgin interior,no doubt partly explainEurope’s marked preference for ships, butrelative prices also played a role.US rail freight rateswerewell below the European norm, while US cabotage rates were not.[21]Among the possible reasons are factor costs, different freight mixes, speeds and journey lengths, railroad land grants, ownership/regulation/competition,higher European rail safety spending, the US ban on foreign crews and on foreign cabotage (compared with Europe’s, largely open, ports), and the failure to invest in Panama as speedily as Europe invested in ship canals.[22] Some of these possible determinants are compatible with marginal social costs differing less than observed market prices.[23] Whatever the reasons, the upshot is clear:the Americandomestic market was glued together primarily by the train, but the First European Integration was driven by the ship.Europe’s long-haul costs per tonne-km were thusbelow what they would have been with aUS ship-train mix.The European imperialists who argued for emulating the transcontinental railroads of the USA and Canada on what Europeans then thought of as their frontier – whether unifying the Tsar’s Empire with the Trans-Siberian railway or consolidating control of Africa with a Cape-to-Cairo line - were deranged dreamers, not transport economists: ships were the efficient option for Vladivostock or Cape Town, as within Europe.[24] The future belonged substantially to sea-routes and already their advantage over land transport was clear.[25]

The major logistical bottleneck of 1900, in which there were also large international differences, was transport by road. At the turn of the century, thiswas, of course, principally by horse-drawn wagon or what in Asia was called the jinrikisha (literally “man-power-vehicle”), but, for most countries,such freights area statistical desert.Road wasunimportant in overall tonne-km terms because, given its cost, it was (before the motorized truck and improved intercity roads) sensiblyavoided, where possible, for long-haul freight(and is therefore omitted from Table 1).[26] Yet,for the same reason, when there was no alternativeand for short-haul trips and final delivery,it added massively to logistical costs.We might reasonably suppose, following the contemporary European stereotype encouraged by American cowboy showmen, that America was much more of a horse-riding and horse-drawn society than western Europe and (despite probable undercounting) the census data on teamsters and horses confirm this. In 1901, for example, there were perhaps 3.3 million horses in Britain, while the number in US cities alone equaled that and in the nation as a whole exceeded 24 million.[27] The huge extent of the American demand for road transport is also shown by thenational production of carriages, wagons, carts and similar mobiles.In 1904 more than 1.7 million carriages and wagons, worth $97 millions at the factory gate,were produced in the USA: a level of unit sales per head of population not to be equaled by the new-fangledUSautomobilesector until the 1920s.[28] Around the turn of the century, the French were only producing 36,000 horse-drawn passenger vehiclesannually, worth 35 million francs ($6.8 millions) and it was reckoned that the total in service was only 1.5 million; the stock of freight vehicles is not recorded.[29]Thompson suggests that in Britainthere were perhaps 702,000 freight carts, vans and wagons in service in 1901 and about the same number of passenger carriages, sothe annual production of horse-drawn vehicles in the USA was of about the same magnitude as the totalstockoutstanding in Britain.[30]TheUKoutput of all “carriages and carts for animal traction” in 1907 was around $6 millions,possibly, given higher British prices, little more than 20,000 units.[31] If that is right, the productionof these road vehiclesfor all private, business, public hire and self-drive (passenger and freight) use wassixty times higherper capitain America than in Britain.Surprisingly, the American road vehicle manufacturing industry, despite the best efforts of Henry Ford and Alfred Sloan, was not able to maintain anything remotely like thisearly lead in the later age of the mass-produced automobile.[32]

The United Statesevidently entered the twentieth centuryas an exceptionally road-using society. Of course, much of the logistical need met by the teamster and his horses in America was performed by the denser European steam railway network.[33]In 1900 the USA’s rail system, at 311,287 km, was understandably a third longer than Europe’s west of Russia, but it served only 76 million people (compared with Europe-ex-Russia’s 285 million, in an area only half the size), so theservices provided were less close to the median producer or consumer and less frequent(most was still single track).[34]The mid-Atlantic regionwas the most densely networked and included four of America’s seven largest cities (New York, Philadelphia,Pittsburgh and Baltimore) as well as major operations of its largest corporations. This region more closely resembled highly urbanized and industrialized north-west Europe and had a land area and track lengthonly a little below the UK’s. Typically in such regions,on both sides of the Atlantic, lines were double-tracked (permittingaround five times thetraffic flow),though British railway managershad to schedulehigher-intensity traffic on theirbusier network: twice as many locomotives serviced a UK population one-and-a-half times larger than the mid-Atlantic region’s.[35]London, alone, had,in the early twentieth century,500 passenger rail stations (where parcels could be consigned or received: the railways competed with the Post Office and road hauliers for this business) and 74 goods depots (for heavier traffic like coal and timber, with 770freight trains aday running among them).[36] Rail companies in Britainwere among the largest importers and owners of horses, but only for the short collection and delivery runsnecessary in such densely rail-served cities.In Germany, too, the railways were more involved in the distribution of parcels and finished manufactures (and less in raw materials) than in the USA.[37]With this well-organized and competitive delivery network (doing work differently divided among express companies, manufacturers’ wagon fleets and self-drive farmers and storekeepers in the USA), Europeanfactories and warehouses could outsourcedistributionto tens of thousands of retailers more cheaply than their American counterparts (who were accordingly more likely to integrate forward to distribution).[38]